Challenges

The challenges of the project are:

Ch1. Defining data-intensive API creation for evolving. Most of the existing works related to API creation deal with the modelling of their static structure and behavioural part without considering their later evolution. Moreover, these works do not take into account the specificities of data-intensive APIs. The complexity of data-intensive API evolution requires new techniques and methods that consider future evolution due to changes in both data sources and requirements.

Ch2. Improving the automatic generation and maintenance of data-intensive API documentation. A complete, accurate and updated documentation is a key factor for the wide use of data-intensive APIs. Currently, the quality, the maintenance and the automatic generation of API documentation are in question. So, new approaches have to be provided to improve the automatic generation and maintenance of data-intensive API documentation, which includes the internal transformations of data.

Ch3. Exploiting available evidence to generate data-intensive API evolution proposals. The observations provided by data-intensive API users (developers of software applications that consume those APIs and end-users of applications using such APIs) as well as the analysis of data-intensive API usage and the software repositories of applications that consume those APIs, are valuable information that needs to be examined in more detail to generate data-intensive API evolution proposals.

Ch4. Defining strategic indicators to evaluate data-intensive API evolution proposals.  When several evolution proposals emerge, it is difficult to choose among them which are the ones to be implemented. The challenge is to define a set of strategic indicators that helps the API provider to evaluate and prioritize the implementation of those proposals.

Ch5. Automating the data-intensive API evolution process. The state of the art in API evolution shows that this process is still quite manual. The mentioned challenges that we will tackle in the data-driven approach we are going to develop, will inevitably lead us to improve the automatization of the data-intensive API evolution process.